Hashrate Growth Explained: What Drives Bitcoin’s Mining Power
Bitcoin's hashrate climbed from a single laptop to briefly top one zettahash in early 2026. Here is what that growth measures, why it matters, and why the network pulled back this summer.
Bitcoin’s hashrate is the single clearest gauge of how much computing muscle sits behind the network, and for most of the asset’s history it has pointed in one direction: up. In January 2026 the seven-day average briefly pushed past one zettahash per second, or 1,000 exahashes, a figure that would have looked absurd even three years earlier, according to CoinWarz tracking data. Then, in a matter of weeks, it fell hard. This explainer walks through what hashrate growth actually measures, the forces that push it higher, and why the summer of 2026 delivered the sharpest reminder in years that the line does not only go up.
What Hashrate Actually Measures
Hashrate is the combined speed at which mining machines on a proof-of-work network guess at solutions to a cryptographic puzzle. Each guess is one hash. Miners run the SHA-256 function over block data billions of times per second, hunting for an output below a target set by the protocol. The first machine to find a valid answer earns the right to add the next block and collect the reward, currently 3.125 BTC plus transaction fees after the April 2024 halving cut the subsidy in half.
Because a single machine now performs trillions of hashes each second, the network total is quoted in exahashes (one quintillion hashes) and, at its 2026 peak, zettahashes (one sextillion). Nobody can measure the figure directly; there is no central meter. Instead, services such as Blockchain.com estimate it from how quickly blocks are actually found against the current difficulty. That is why published hashrate numbers wobble day to day, and why analysts lean on seven-day averages to smooth the noise.
From Megahashes to a Zettahash: The Growth Curve
When Satoshi Nakamoto mined the genesis block in 2009, a single laptop CPU carried the entire network. The arc since then tracks the shift from CPUs to GPUs, then to field-programmable chips, and finally to purpose-built ASICs that do nothing but hash. Each leap multiplied output by orders of magnitude. The table below shows approximate year-end network hashrate drawn from public chart data; treat the values as round estimates rather than precise readings.
| Year | Approx. network hashrate | Context |
|---|---|---|
| 2016 | ~2.5 EH/s | Early ASIC era |
| 2018 | ~40 EH/s | Post-2017 bull run buildout |
| 2020 | ~150 EH/s | Pre-China-ban capacity forming |
| 2021 | ~175 EH/s | Recovery after China’s mining ban |
| 2022 | ~265 EH/s | Bear market, yet capacity rose |
| 2023 | ~500 EH/s | US industrial scale-up |
| 2024 | ~800 EH/s | First full year after the 2024 halving |
| Jan 2026 | >1,000 EH/s | Brief cross above one zettahash |
| Jul 2026 | ~890 EH/s | After a spring and summer pullback |
The pattern is striking: even through deep bear markets, capacity kept climbing. Cheaper, more efficient hardware plus a steady inflow of institutional capital meant that falling prices rarely translated into falling hashrate for long. That resilience is exactly what makes the 2026 reversal worth studying.
The Difficulty Adjustment Keeps Blocks on Schedule
Left alone, more hashrate would mean faster blocks. Bitcoin prevents that with a difficulty adjustment every 2,016 blocks, roughly every two weeks, that retargets the puzzle so blocks keep arriving about every ten minutes. When hashrate rises, difficulty rises to match; when machines switch off, difficulty falls at the next retarget.
Difficulty is therefore a lagging mirror of hashrate. It set an all-time high near 144 trillion in February 2026, per CoinWarz data, the biggest single upward jump the network had ever recorded, before the summer sell-off forced a retreat. The design is deliberately self-correcting, holding the ten-minute cadence whether the network is booming or bleeding capacity.
What Pushes Hashrate Higher
Three forces drive long-run hashrate growth. The first is chip efficiency. Modern ASICs are rated in joules per terahash (J/TH); the lower the number, the more hashing you get per unit of electricity. In 2026 the frontier crossed a symbolic line when hydro-cooled units broke below 10 J/TH, a threshold that reshapes which older machines can still turn a profit.
| Machine (2026) | Efficiency | Rated hashrate |
|---|---|---|
| Antminer S23 Hydro | ~9.5 J/TH | First sub-10 J/TH unit |
| Antminer S21 XP Hydro | ~12.0 J/TH | 473 TH/s |
| Antminer S21 XP (air) | ~13.5 J/TH | 270 TH/s |
| Avalon A15 Pro | ~15.0 J/TH | 221 TH/s |
The second force is energy. Miners chase the cheapest reliable power on the planet, whether that is stranded gas in Texas, hydro in the Pacific Northwest, or curtailed wind. The third is capital: publicly listed operators can raise money on equity markets and pour it into tens of thousands of rigs at once. Location matters as much as headline price, since a rig that pays for itself on a cheap Texas contract can bleed money on a costlier grid. Put the three together and, in normal conditions, you get a machine that turns falling hardware costs and cheap electricity into ever more hashing.
Hashprice: Why Growth Squeezes Margins
Here is the catch that newcomers miss: hashrate growth is bad news for the average miner’s revenue. The block reward is fixed per ten-minute interval, so when more machines join, each terahash claims a thinner slice of the same pie. The industry tracks this through hashprice, the estimated daily revenue per terahash, published by Hashrate Index and mirrored by The Block.
Hashprice is a tug of war between three variables: the Bitcoin price in USD, transaction fees, and difficulty. Rising prices lift it; rising difficulty drags it down. Since the 2024 halving cut the subsidy, miners have had to run leaner just to stand still, which is why efficiency and cheap power contracts, not raw scale, now separate the winners from the casualties.
The 2026 Pullback: When Hashrate Falls
The first half of 2026 showed the other side of the curve. After the January peak above one zettahash, Bitcoin’s price slid from roughly $81,000 toward the low $60,000s by June, briefly touching $61,165 on CoinGecko pricing. That roughly 15% monthly drop pushed a swath of high-cost machines below breakeven.
The response showed up in the data fast. Roughly 90 EH/s of capacity powered down, and the network booked a difficulty cut of about 10%, from 138.96 trillion to 124.93 trillion, one of the largest downward adjustments of the year. Analysts at JPMorgan, cited by The Block, pegged all-in production cost well above the spot price, and the CoinShares Q1 2026 mining report had already flagged 15% to 20% of the global fleet as unprofitable. Several operators had also sold tens of thousands of coins from their own treasuries during the quarter to cover costs, deepening the squeeze. Growth, it turns out, is conditional rather than guaranteed.
Security: What a Bigger Network Buys
Hashrate is also Bitcoin’s immune system. To rewrite recent history or double-spend, an attacker would need to out-hash the honest network, the so-called 51% attack. The more hashrate online, the more machines and electricity that attacker must marshal, and the more absurd the economics become. At roughly 900 EH/s, replicating even a fraction of the network would cost billions in hardware alone, before a single watt of power. Put differently, the honest network’s hardware doubles as its moat, and years of growth have widened that moat faster than any challenger could dig.
This is the quiet dividend of years of growth. It also cuts the other way: when hashrate falls sharply, as it did this summer, the theoretical cost of an attack falls with it. For a network as large as Bitcoin the risk stays remote, but for smaller SHA-256 chains that share the same mining hardware, a hashrate exodus can be genuinely dangerous.
Where the Machines Live, and Why the SEC Cares
After China outlawed mining in 2021, the industry’s center of gravity moved to North America. The United States now hosts the largest share of global hashrate, per the Cambridge Centre for Alternative Finance mining map, and the US Energy Information Administration tracks the sector’s electricity draw closely enough to publish its own estimates.
That US concentration hands a regulator a starring role. The biggest miners are public companies, and their hashrate is now a disclosed operating metric. MARA Holdings reported about 72 EH/s of energized hashrate at the end of the first quarter, while CleanSpark crossed 50 EH/s built entirely on American sites, figures compiled by trackers such as Bitcoin Mining Stock. Because these firms trade on US exchanges, they file quarterly production and risk disclosures with the Securities and Exchange Commission, searchable through the SEC’s EDGAR database. For investors, those filings, not a raw chart, are the authoritative record of who controls the hashrate.
The AI Pivot and What to Watch Next
The defining shift of 2026 is that some of the largest operators no longer want all their power pointed at Bitcoin. Hosting artificial intelligence and high-performance computing workloads can pay more per megawatt than mining, and it does not care about hashprice. A growing list of listed miners has redirected capacity toward AI data centers, with at least one firm dropping the word Bitcoin from its name outright.
For hashrate, this creates a new ceiling. When mining margins compress, that power can now walk out the door to a higher bidder instead of sitting idle waiting for the next rally. Three signals are worth watching from here:
- Hashprice, the clearest real-time gauge of miner profitability.
- The pace at which sub-10 J/TH rigs displace older hardware.
- How much energy the public miners keep aimed at Bitcoin rather than AI.
Hashrate growth is not dead, but for the first time in years it faces a serious competitor for the same electrons. The long climb that carried Bitcoin from one laptop to a zettahash now runs through a boardroom decision about where the cheapest power should go.
By the HOGE Wire Mining and Staking Desk. Written for information only; nothing here is investment advice.